Imagine giving Claude a bunch of files and asking for a CSV or JSON output. If you have tried this, you know both the system limitations (number of files, type of inputs, cost, latency) but also the human-facing challenge of having no way to validate the results quickly. We solve both. We help tech teams simplify their unstructured data ETL, and loop in business experts for the definitions and for instant validation.
Here is a video with a few use cases: https://www.youtube.com/watch?v=dbRllnnh47w
Parsewise in the words of someone coming to us: ”I need to extract information from insurance policy PDFs, phone calls that have been transcribed, emails, etc. I am NOT looking for something that would just extract data point by data point, page by page into a structured well-defined schema but more something more agentic that can understand that information might be across documents and that it should reason over what to extract.”
We started the company based on a decade of experience (and pain) in complex data transformation and data analysis / synthesis. Greg was building both classical ETL and implemented AI workflows at Palantir. At Bain, Max did highly complex data analysis in the financial sector, similar to many of our customers.
Parsewise works by taking in a bucket of data (think hundreds or thousands of pdfs, excels etc.), and outputting schema compliant data where every single value is traceable down to word level citations across multiple documents in the bucket. We provide API customers with ways to show the lineage in their own applications, or they can use our platform for internal operations. At the core of the data processing we have self-improving agent definitions. They define the acceptable sources, the logic for resolving or combining values, and the rule for highlighting uncertainty to the end user.
The underlying tech is model and cloud agnostic and can be deployed in private networks. We have seen the best results with Gemini models for visual reasoning, achieving SOTA (beating Claude Fable) on the strongest grounded reasoning benchmark we have found (Databricks OfficeQA). Notably, we focused more on the “human harness” rather than the model harness, leaning into the actual friction we saw in uptake, which is around verifiability. That means optimizing the time and clicks required to trust the outcomes. We use vLLMs for parsing, and then we use small models for efficient large scale exhaustive search. Unlike RAG, we do not sample; instead, we exhaustively find all relevant values for a given query. We use larger models for decision making around resolutions and flagging inconsistencies to users.
This exhaustiveness and explicit value sourcing is unique to our platform, and it goes beyond the first step of data parsing that many existing providers cover.
We would love to welcome builders and tinkerers to try Parsewise on your complex document challenges. We have a ton of ideas on how we can expand the product and make it better, but would appreciate feedback and ideas from the community!
At first, fairly generic templated definitions sort of work, but what we've seen is that over time data comes up that is out of distribution, and there was no explicit instruction on how to deal with it. In such cases we tend to flag this and offer suggestions to the users on how they can improve the specificity of agents.
Another structure we have seen play out is having a manager review ratings and feedback comments from their team and updating the definitions accordingly over time (where we offer them the capability to see results of before and after side -by-side for all existing data as well, so they are more confident in the change before committing).
The amount of work is dependent on how good the initial definitions are and how complex the use case is (and how much it evolves - new data sources etc). A bit of an unsatisfying answer but it can be anywhere between a few hours one off or a couple of minutes per day on an ongoing basis.
There seems to be an insane amount of competition in the "Intelligent Document Processing" market, like for instance parseur, whose founder is often on HN himself.
What do you think sets you apart from competition like : 1) Mistral document AI : depending on the model, it looks way cheaper than yours, OCR model pricing ranges from 0.001 to 0.004 EUR / page and they have structured output wired in the OCR API if needed (things then get fed to one of their LLMs) + EU-based and GDPR ready 2) parseur / rossum / docsumo / nanonets (which is YC 2017) ?
I understand what they are trying to do, but to me it feels like the moment when MongoDB entered the database space, with semi-structured, "flexible" storage format. It has its uses, for prototyping mostly.
But in high-volume, production workloads, giving a structure to the data you extract (what Parseur does through defining the Fields in your Mailbox, basically giving your output data a schema) adds a ton of value, and the larger the dataset, the truer it is.
Usually, you start by defining where you want your data to go, and which structure it should have, before working backwards from here and starting to extract the data. This is the key to automating your document workflow.
1. We are working with the assumption that OCR is (or soon will be) solved at super low prices.
So if we have the extracted data, what can we do with it? Where we see Parsewise making a difference is for use cases that span across documents. I.e. if you are extracting the same 5 fields from every invoice, there are lots of solutions as you listed (+ reducto etc). However, once you have a set of documents (e.g. an entire mortgage application package) and you are trying to get a structured response out, then your option is either an LLM API (if things fit into context and you are okay with limited citations), or building a pipeline with LLMs. I posted it in another comment but an example of trawling through 90k pages is here: https://www.parsewise.ai/officeqa-sota
2. While we rely on LLMs, the outcomes will be non-deterministic, so the bottleneck is and will remain the human verification (that is for somewhat complex use cases). The architecture that we have built is optimizing for the human reviewer to provide as granular values and citations as possible. This is either through our platform, or API clients.
This isn't a critique of your product. It's just that the a beige-orange theme, the pill components, and the left-border highlight give me that visceral reaction as reading a paragraph littered with em dashes and "not X but Y." It makes me take you less seriously.
Cool demo otherwise.
https://news.ycombinator.com/newsguidelines.html
And I do get genuine joy from helping our users, so love it is:)
https://news.ycombinator.com/newsguidelines.html